Can I use WaveSpeed LLM models with Codex?

Yes. Use the OpenAI-compatible base URL, your WaveSpeed API key, and a supported model id in the Codex provider configuration.

Does this work with Claude Code, OpenClaw, OpenCode and Hermes?

Yes. The setup page provides copy-ready environment variables and configuration examples for Claude Code, OpenClaw, OpenCode, Hermes and OpenAI-compatible agents.

Where do I get my API key?

Create or copy an access key from the WaveSpeed AI Access Key page, then paste it into your local agent environment.

Agent setup

Connect your local AI agent to WaveSpeed LLM

Pick an agent, choose a WaveSpeed model, copy the right local config file, and route requests through the protocol that agent expects.

API key Test model

Config file

~/.codex/config.toml

Protocol

OpenAI Responses API

Endpoint

/v1/responses

Quick start

Create a WaveSpeed access key

Export the required key locally

Paste config into~/.codex/config.toml

Run codex and check Logs

Model

OpenAI-compatible chat completions

Usage appears in Dashboard and Logs

Switch models without rewriting agent code

Edit ~/.codex/config.toml

# File: ~/.codex/config.toml

model = "openai/gpt-5.6-sol"
model_provider = "wavespeed"

[model_providers.wavespeed]
name = "WaveSpeed AI"
base_url = "https://llm.wavespeed.ai/v1"
env_key = "WAVESPEED_API_KEY"
wire_api = "responses"

Do not replace the variable name

WAVESPEED_API_KEY is the environment variable name that the agent reads. Keep this name in the config file, and put your real API key in your shell instead.

Temporary for this terminal

Works until you close this terminal window.

export WAVESPEED_API_KEY="wsk-..."

Keep it after restart

Writes the key into your shell profile so new terminals can read it.

echo 'export WAVESPEED_API_KEY="wsk-..."' >> ~/.zshrc
source ~/.zshrc

Connect your local AI agent to WaveSpeed LLM

Quick start

Use WaveSpeed AI as an LLM agent gateway

Agent setup guides